Medical students’ perceptions of professional mission in an AI-driven healthcare future: a text mining analysis of reflective essays in Japan

Abstract

Objective

As artificial intelligence (AI) technologies advance rapidly, their integration into healthcare is transforming the clinical practice landscape. This study aimed to evaluate how second-year medical students perceive their professional mission in an AI-integrated medical future, through a structured essay task, using text-mining analysis to identify emerging themes and attitudes.

Methods

A total of 105 second-year medical students at Kagawa university in Japan completed an essay titled “What is your mission in the AI-driven medical world?”. Responses were analyzed using KH Coder for frequency analysis, multidimensional scaling, and co-occurrence network mapping. Participants provided verbal informed consent and student anonymity was ensured.

Results

The most frequently used terms were medical, consider, think, doctor, AI, human, and patient. Three thematic clusters emerged: (1) career design, (2) AI and medicine, and (3) AI and human. Co-occurrence analysis revealed strong associations between “medical” and both “consider” and “patient”, while “patient” was linked to both “AI” and “human”, indicating thoughtful reflection on technology’s impact on patient care.

Conclusion

Second-year medical students in Japan demonstrated critical engagement with the concept of mission formation in the context of AI in healthcare. Their essays reflected a balance between optimism for technological advancement and concern for preserving human-centered care. These findings highlight the importance of implementing systematic career education and future-oriented thinking that is aligned with the characteristics of Generation Z learners.

Introduction

In Japan, individuals can enter medical school directly after graduating from high school at the age of 18 years, assuming that they have progressed through their education without repeating academic years or taking time off for university entrance preparation^{1, 2)}. After completing six years of undergraduate medical education and passing the national medical licensing examination, most graduates begin a compulsory two-year clinical training program, usually between the ages of 24 and 27. Following this, many pursue specialization in areas such as internal medicine, surgery, obstetrics, or critical care, which typically involves an additional 4–5 years of postgraduate training. Certification in a medical specialty is generally attained approximately 7–11 years after graduating, marking a transition to a lifelong process of continued professional development³⁾. In the Japanese medical context, it is common for physicians to pursue doctoral degrees after acquiring clinical research experience. However, the noticeable decline in the number of doctors obtaining PhDs has emerged as a critical concern, closely associated with diminishing research output and capabilities in the medical sector⁴⁾.

The recent emergence of advanced artificial intelligence (AI) technologies, particularly those based on deep learning, has introduced what is often referred to as the third-generation AI. These technologies are beginning to play a transformative role in healthcare⁵⁾. Unlike earlier forms of AI, these systems can mimic human cognitive functions, such as recognizing patterns, drawing inferences, and learning from past data, enabling their use in diagnostic support, treatment planning, and biomedical research^{6, 7)}. For instance, AI algorithms are now capable of interpreting medical imaging data to identify abnormalities, thereby aiding clinicians in making informed decisions⁸⁾. Moreover, AI contributes to streamlining healthcare operations by efficiently processing large datasets, thereby enhancing system-wide productivity⁹⁾. Tools, such as AI-powered electronic health records, can automatically summarize patient histories, allowing healthcare professionals to make quicker and more informed judgments¹⁰⁾.

Given the increasing integration of AI into medical practice, it is imperative that medical curricula adapt accordingly. This includes the incorporation of training in AI ethics, limitations, and effective utilization. Equipping future physicians with the necessary knowledge and competencies in these areas is crucial for the responsible application of AI in clinical settings.

In our previous work, we emphasized the importance of fostering career awareness among medical students in Japan^{11, 12)}. Many students lack a clear understanding of their long-term professional paths. To address this issue, we implemented career design simulations targeting fourth-year students, which proved beneficial in nurturing their ability to think about career development and lifelong learning¹³⁾. Although these simulations were effective for senior students, we identified the need for a structured approach to support younger students as well. Specifically, introducing a form of career education that emphasizes personal mission development is expected to foster a deeper reflection on professional goals. Based on this, we introduced “mission consideration” classes for students in the earlier years of medical school.

As AI technologies rapidly advance and become increasingly integrated into healthcare, clinical practice is undergoing a significant transformation. Considering this shift, we evaluated how future physicians perceive their professional roles. This study aimed to evaluate how second-year medical students perceive their professional missions in an AI-integrated medical future. Apropos this, we implemented an educational intervention for second-year medical students centered on writing their medical mission. Their written responses were quantitatively analyzed using text mining methods, including keyword extraction, multidimensional scaling, and co-occurrence network analysis to uncover thematic patterns and insights into their perspectives on practicing medicine in an AI-integrated future.

Material and Methods

Ethical considerations

We consulted the Institutional Review Board of the Faculty of Medicine at Kagawa University and determined that no application was necessary, provided that anonymity was maintained and individual academic records remained unaffected. Verbal informed consent was obtained from the participants by medical faculty members, and student clerks observed the consent procedure prior to the survey. All the participants were informed in advance of the study’s objectives and procedures, and an assurance of anonymity was clearly communicated. Furthermore, students were notified that they could withdraw from the study within one week of survey completion, with a clear explanation that doing so would not influence their academic evaluations. Notably, the participants were all second-year medical students in Japan, who are uniformly aged ≥18 years.

Study population and setting

In Japan, medical education generally spans six years, beginning immediately after high school, for students who have successfully passed competitive entrance examinations. At Kagawa University’s Faculty of Medicine—consistent with the standard curricula nationwide—students receive extensive instruction in both basic and clinical sciences, along with hands-on skills training prior to starting their clinical clerkships, which typically begin in the fourth year^{14, 15)}.

Participants were recruited from Kagawa University’s Faculty of Medicine, the only medical school in Kagawa Prefecture, which is situated in the northeastern part of Japan’s Shikoku region and has an estimated population of approximately 910,000 people.

Essay on their mission

On May 8, 2024, we conducted classes on “Professionalism and Behavioral Sciences” for second-year medical students. We then assigned an essay task posing the following question: “What is your mission in the AI-driven medical world?”. The length of the essay was set to 800–1,500 Japanese words, to be completed in the maximum time of 90 min. Students submitted their essays during the class via the online communication system WebClass^TM (Japan Data Pacific, Tokyo, Japan).

Text mining methodology

Text mining offers a systematic and unbiased approach to analyzing large volumes of student-written essays. This enables researchers to uncover underlying patterns in learning contexts and construct models that reflect students’ educational experiences^{16, 17)}. In the present study, we used KH Coder 3.02 (https://khcoder.net/), a free multilingual software tool for text analysis, developed by Koichi Higuchi at Ritsumeikan University¹⁸⁾.

To begin the analysis, all the essays were compiled into a single document. A frequency analysis was conducted on this combined text to identify commonly occurring words. Words that shared similar meanings or referred to the same concept, such as “clinical” and “medicine”, were grouped together to streamline the analysis. The focus was placed specifically on extracting frequently used nouns as these tend to represent key concepts within the texts.

Thereafter, multidimensional scaling was used to examine the relationships among the extracted terms. This statistical technique enables the visualization of clusters, where words that frequently appear together in similar contexts are positioned closer to one another, whereas less related terms appear farther apart. This method helped in revealing patterns in how students expressed related ideas.

To further investigate the interrelationships among the high-frequency terms, co-occurrence network maps were created based on a co-occurrence index. This index measures how often certain key terms—defined as frequently appearing words—appear near each other in a text. The strength of these relationships, also known as associative strength or inclusion, provides insights into the semantic structure of the content. The co-occurrence index values range from 0 (no co-occurrence) to 1 (perfect co-occurrence), and in this study, the Jaccard similarity co-efficient was applied to calculate these values. The resulting network maps offer a visual depiction of the associations among the key terms across the dataset. The lines connecting words in the map are annotated with numerical values that indicate the degree of association, illustrating the frequency and closeness of their co-occurrence.

Patient and public involvement

None.

Results

In total, 105 of the 115 second-year medical students responded to the survey (response rate: 91.3%). None of the students asked to exclude their essay from the analysis. Brute-force word extraction of the combined essays (total: 15,358 words; unique words: 1,674) identified seven nouns that occurred most frequently: medical (151), consider (148), think (137), doctor (114), AI (113), human (85), and patient (84).

The results of the multidimensional scaling are presented in Figure 1, which shows that the relative distance between each word and the axis has small significance. There was a total of three clusters, which could be divided into the following three major clusters: “Attitude toward career design”, “Attitude toward AI and Medicine”, and “Attitude toward AI and human”.

Figure 1.

Multidimensional scaling. (a) Japanese, (b) English.

Co-occurrence (i.e., the proximity of keywords to other high-frequency words) is shown in Figure 2. Co-occurrence network calculations revealed that the word “medical” had a high degree of correlation with “consider” and “patient”. The word “patient” was correlated with not only “medical” but also with “AI” and “human”. The word “doctor” correlated with the word “consider” but not with “think’ (Table 1).

Figure 2.

Co-occurrence network map with degree values, with the seven most frequently-occurring terms clustered together. (a) Japanese, (b) English. Co-efficient was calculated as the Jaccard similarity index.

Table 1. Degrees of co-occurrence between keywords based on keyword map (empty cells: <0.40).

	Medical	Consider	Think	Doctor	AI	Human	Patient
Medical	N.A.	0.52					0.52
Consider		N.A.	0.50	0.46
Think			N.A.
Doctor				N.A.
AI					N.A.		0.49
Human						N.A.	0.40
Patient							N.A.

N.A.: not applicable.

Discussion

This study analyzed essays written by second-year medical students to explore their perceptions and attitudes toward medicine and artificial intelligence (AI). These findings are likely reflective of broader cohort-level perspectives.

Three major thematic clusters—“career design”, “AI and medicine”, and “AI and human”—emerged from the analysis, indicating that students are beginning to develop nuanced perspectives that encompass both their professional aspirations and ethical considerations related to AI in healthcare. The strong associations between the term “medical” with the words “consider” and “patient” imply that students are reflecting on the essence of medicine with a patient-centered mindset. Furthermore, the word “patient” was associated not only with “medical” but also with “AI” and “human”, suggesting that students are beginning to grapple with how AI might reshape the patient-physician relationship and the delivery of humanistic care.

Interestingly, the word “doctor” was linked to “consider” but not to “think”, which may suggest a more pragmatic or decision-oriented stance toward their future roles. In this report, “think” was used when expressing subjective opinions or feelings superficially, whereas “consider” was used to indicate judgments made after evaluation and careful thought. The frequent use of reflective verbs such as “consider” suggests active engagement and contemplation regarding their future roles in medicine and the integration of AI into clinical practice. This implies that students may view AI primarily as a tool for improving diagnostics and treatment while maintaining the belief that it is the human physician who ultimately bears the responsibility of treating the patient as a whole person. Thus, it can be inferred that while students are aware of AI’s growing relevance, they may not perceive it as fundamentally altering their career trajectories.

In this context, consideration of one’s personal mission becomes crucial in medical education. A clearly articulated mission provides students with a strong sense of direction and intrinsic motivation, extending beyond technical proficiency to encompass their values, aspirations, and broader impacts they wish to make. This helps guide important career decisions such as choosing a specialty, selecting research topics, and shaping clinical approaches^{19, 20)}.

A mission-driven mindset also fosters long-term resilience, allowing students to maintain their focus and purpose throughout the often demanding and extended process of medical education. Importantly, students with a defined mission are more likely to approach patient care with empathy and compassion, contributing to improved healthcare outcomes and stronger doctor-patient relationships²¹⁾.

Moreover, a personal mission helps develop a strong professional identity. It encourages ethical decision-making, a sense of responsibility within the healthcare system, and accountability as future physicians²²⁾. In a broader context, this inspires medical students to pursue contributions beyond individual gains, such as community engagement and socially impactful research. It also cultivates a commitment to lifelong learning, an essential quality in a field characterized by rapid advancements²³⁾.

Notably, the text mining results showed that only a small cluster expressed views that truly centered on the patient as a human being. This may indicate an underdeveloped understanding of the concept of mission among the current students. From the perspective of self-regulated learning, this highlights the need to actively cultivate mission awareness throughout medical training. Lifelong learning, grounded in a well-formed mission, is essential for maintaining high standards in the evolving landscape of medicine²⁴⁾.

The relevance of this discussion is further underscored by generational shifts in medical education. Most current medical students belong to the Generation Z, while their instructors are typically from Generations X and Y. Generation Z students are digital natives who value creativity, diversity, and immediate access to information. While adept at navigating digital tools and online communication, they may face challenges in forming deep, reflective judgments in a fast-paced information environment²⁵⁾. Given these generational characteristics, traditional educational approaches, such as inspirational lectures that highlight medical role models, may be less effective than in the past. Therefore, systematic career education that aligns with Generation Z’s learning style and strengths is vital^{26, 27)}. The revised Model Core Curriculum for Medical Education in Japan has begun to address this issue by emphasizing career diversity among physicians. Starting a systematic career design education at the undergraduate level can enhance students’ self-regulated learning and support their development into well-rounded professionals²⁸⁾.

At our university, we are implementing a career design curriculum based on a conceptual framework centered on three key elements: “knowing oneself”, “understanding what one wants to gain”, and “considering what one can contribute to the society”. This framework is integrated into early stage courses, including those on data science, AI, interprofessional collaboration, and professionalism. As students progress to clinical clerkship, they build upon these foundations by interacting with various supervising physicians, thereby gaining practical career design skills²⁹⁾. This structured approach to career education is particularly relevant in supporting the development of diverse medical careers in the digital age, where adaptability and reflective practices are increasingly necessary.

This study has several limitations. First, the data were collected from a single institution, which may limit the generalizability of the findings to medical schools in other cultural or educational contexts. Second, although the sample included both male and female students, gender differences in perceptions of mission and career design were not analyzed. This represents a potentially valuable avenue for future research^{30, 31)}. Third, while this study employed text mining as a qualitative analytical method, future studies could complement it with quantitative approaches such as visual analog scales or Likert scales to more precisely assess students’ awareness or attitudes toward mission formation and career design³²⁾. Finally, the purpose of text mining is to extract patterns and insights from large volumes of textual data, and is not intended to focus on individual responses. Therefore, in future studies, it may be necessary to explore ways to extract students’ perceived mission from their reports.

Conclusion

This study analyzed how second-year medical students in Japan perceive their professional mission in an AI-driven healthcare environment. Text mining revealed that while students recognize the role of AI, their focus remains on patient-centered care, reflecting an emerging but still developing understanding of professional mission. The limited emphasis on viewing patients as whole human beings suggests the need to foster deeper mission awareness. Introducing structured, mission-oriented career education early in medical training is essential, particularly for Generation Z students, who thrive with self-directed and reflective learning approaches. As AI continues to reshape medicine, equipping students with both technological literacy and a clear professional identity will be the key to ensuring responsible humanistic medical practices. Future research should further explore how mission development influences career trajectories and patient care.

Funding

This work was supported by the Establishing Bases for Fostering Medical Personnel in the Post-COVID Era Project by the Japanese Ministry of Education, Culture, Sports, Science and Technology.

Conflict of interests

The authors have no affiliation with the manufacturers of any devices described in the manuscript and declare no financial interest in relation to the material described in the manuscript.

Ethics consideration and consent to participate

We consulted the Institutional Review Board of the Faculty of Medicine at Kagawa University and no application for approval was deemed necessary, provided that anonymity was maintained and individual academic records remained unaffected.

Consent for publication

All authors consent to the publication of this article in the Journal of Rural Medicine.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Author contributions

N.K. performed the study, statistical analysis, and wrote the manuscript; M.Y. prepared the manuscript, provided critical comments, and approved the final version. All the authors have read and approved the final version of this manuscript.